Accelerating Masked Image Generation by Learning Latent Controlled Dynamics

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhu, Kaiwen, Zeng, Quansheng, Pu, Yuandong, Cao, Shuo, Li, Xiaohui, Xin, Yi, Qin, Qi, Li, Jiayang, Qiao, Yu, Gu, Jinjin, Liu, Yihao
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911473767284736
author Zhu, Kaiwen
Zeng, Quansheng
Pu, Yuandong
Cao, Shuo
Li, Xiaohui
Xin, Yi
Qin, Qi
Li, Jiayang
Qiao, Yu
Gu, Jinjin
Liu, Yihao
author_facet Zhu, Kaiwen
Zeng, Quansheng
Pu, Yuandong
Cao, Shuo
Li, Xiaohui
Xin, Yi
Qin, Qi
Li, Jiayang
Qiao, Yu
Gu, Jinjin
Liu, Yihao
contents Masked Image Generation Models (MIGMs) have achieved great success, yet their efficiency is hampered by the multiple steps of bi-directional attention. In fact, there exists notable redundancy in their computation: when sampling discrete tokens, the rich semantics contained in the continuous features are lost. Some existing works attempt to cache the features to approximate future features. However, they exhibit considerable approximation error under aggressive acceleration rates. We attribute this to their limited expressivity and the failure to account for sampling information. To fill this gap, we propose to learn a lightweight model that incorporates both previous features and sampled tokens, and regresses the average velocity field of feature evolution. The model has moderate complexity that suffices to capture the subtle dynamics while keeping lightweight compared to the original base model. We apply our method, MIGM-Shortcut, to two representative MIGM architectures and tasks. In particular, on the state-of-the-art Lumina-DiMOO, it achieves over 4x acceleration of text-to-image generation while maintaining quality, significantly pushing the Pareto frontier of masked image generation. The code and model weights are available at https://github.com/Kaiwen-Zhu/MIGM-Shortcut.
format Preprint
id arxiv_https___arxiv_org_abs_2602_23996
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Accelerating Masked Image Generation by Learning Latent Controlled Dynamics
Zhu, Kaiwen
Zeng, Quansheng
Pu, Yuandong
Cao, Shuo
Li, Xiaohui
Xin, Yi
Qin, Qi
Li, Jiayang
Qiao, Yu
Gu, Jinjin
Liu, Yihao
Computer Vision and Pattern Recognition
Masked Image Generation Models (MIGMs) have achieved great success, yet their efficiency is hampered by the multiple steps of bi-directional attention. In fact, there exists notable redundancy in their computation: when sampling discrete tokens, the rich semantics contained in the continuous features are lost. Some existing works attempt to cache the features to approximate future features. However, they exhibit considerable approximation error under aggressive acceleration rates. We attribute this to their limited expressivity and the failure to account for sampling information. To fill this gap, we propose to learn a lightweight model that incorporates both previous features and sampled tokens, and regresses the average velocity field of feature evolution. The model has moderate complexity that suffices to capture the subtle dynamics while keeping lightweight compared to the original base model. We apply our method, MIGM-Shortcut, to two representative MIGM architectures and tasks. In particular, on the state-of-the-art Lumina-DiMOO, it achieves over 4x acceleration of text-to-image generation while maintaining quality, significantly pushing the Pareto frontier of masked image generation. The code and model weights are available at https://github.com/Kaiwen-Zhu/MIGM-Shortcut.
title Accelerating Masked Image Generation by Learning Latent Controlled Dynamics
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2602.23996